Voxtral Text to Speech ComfyUI workflow for Videcool

The Voxtral Text to Speech workflow in Videcool provides a fast and practical way to generate speech from text prompts. Designed for simplicity, clarity, and creative control, this workflow is served by ComfyUI and uses a Voxtral TTS pipeline to turn written text into synthesized speech audio that can be saved and reused in Videcool projects.

What can this ComfyUI workflow do?

In short: Text-to-speech synthesis.

This workflow takes text input and synthesizes new speech audio using a selected preset voice. It loads the Voxtral TTS model, applies generation settings such as voice preset, seed, frame limit, CFG alpha, noise scale, and Euler step count, then produces an audio result that is saved as an MP3 file. The included workflow is optimized for simple local text-to-speech generation without requiring a reference audio sample or voice cloning step.

Example usage in Videcool

Figure 1 - Voxtral Text to Speech ComfyUI workflow in Videcool

Download the ComfyUI workflow

Download ComfyUI Workflow file: voxtral_text_to_speech-api.json

Image of the ComfyUI workflow

This figure provides a visual overview of the workflow layout inside ComfyUI. Each node is placed in logical order to establish a clean and efficient text-to-speech pipeline. The structure makes it easy to understand how the Voxtral model loader, the speech generation node, and the MP3 audio output node interact. Users can modify the text prompt, voice preset, and synthesis parameters to create custom speech variations or integrate text-to-speech into larger audio or video production pipelines.

Figure 2 - Voxtral Text to Speech workflow

Installation steps

Step 1: Install the custom node package that provides the FL_VoxtralTTS_ModelLoader and FL_VoxtralTTS_Generate nodes in ComfyUI so the workflow can load and run the Voxtral TTS model.
Step 2: Install or place the required Voxtral text-to-speech model files according to the custom node's installation instructions so the model loader node can initialize successfully. The workflow itself uses device="auto", dtype="float32", and force_reload=false.
Step 3: Download the voxtral_text_to_speech-api.json workflow file into your home directory.
Step 4: Restart ComfyUI so the custom node and model files are recognized.
Step 5: Open the ComfyUI graphical user interface (ComfyUI GUI).
Step 6: Load the voxtral_text_to_speech-api.json workflow in the ComfyUI GUI.
Step 7: Enter the text you want the model to speak in the text field of the FL Voxtral TTS Generate node.
Step 8: Choose a voice preset such as casual_male, then optionally adjust seed, max_frames, cfg_alpha, noise_scale, and euler_steps depending on the speech quality and variation you want. The attached workflow uses defaults of seed 42, max_frames 2048, cfg_alpha 1.2, noise_scale 1, and euler_steps 8.
Step 9: Run the workflow to synthesize speech and save the output as an MP3 file through the Save Audio (MP3) node. The default filename prefix in the workflow is audio/ComfyUI and the MP3 quality is set to V0.
Step 10: Open Videcool in your browser, select the Text to Speech tool, and use the generated audio in your video or narration projects.

Installation video

The workflow requires only a text prompt and a few basic parameter adjustments to begin generating speech. After loading the JSON file, users can enter the text to be spoken, choose a voice preset, and adjust generation settings such as seed, frame limit, CFG alpha, noise scale, and Euler steps. Once executed, the model generates speech audio and saves it as an MP3 file that can be used across other Videcool tools.

Prerequisites

To run the workflow correctly, install the ComfyUI custom nodes that provide the Voxtral TTS loader and generation nodes, then install the required Voxtral model files according to that node package. These components ensure the workflow can process text input, generate speech audio, and save the result as an MP3 file. Proper installation into your ComfyUI custom node and model directories is essential before running the workflow.

Custom node classes required by the workflow
FL_VoxtralTTS_ModelLoader
FL_VoxtralTTS_Generate
SaveAudioMP3

How to use this workflow in Videcool

Videcool integrates seamlessly with ComfyUI, allowing users to generate speech directly without managing the underlying node graph. After importing the workflow file, simply enter your text, choose the desired voice preset, and click generate. The system handles the backend interaction with ComfyUI, making text-to-speech intuitive and accessible even for users who do not want to work directly with ComfyUI nodes.

ComfyUI nodes used

This workflow uses the following nodes. Each node performs a specific role, such as loading the Voxtral text-to-speech model, processing text input, synthesizing speech audio, and finally saving the output as an MP3 file. Together they create a compact and reliable pipeline that can be easily extended or customized.

FL Voxtral TTS Model Loader
FL Voxtral TTS Generate
Save Audio (MP3)

Base AI model

This workflow is built on a Voxtral text-to-speech pipeline exposed in ComfyUI through the FL_VoxtralTTS_ModelLoader and FL_VoxtralTTS_Generate nodes. The attached workflow is configured for automatic device selection, float32 model loading, and direct speech generation from text using a preset voice rather than voice cloning from a reference recording.

Speech generation quality and parameters

Speech quality and runtime behavior in this workflow depend on the chosen text, voice preset, and generation parameters. The attached workflow uses the casual_male voice preset with seed 42, max_frames 2048, cfg_alpha 1.2, noise_scale 1, and euler_steps 8, which gives users a practical starting configuration for general text-to-speech tasks. Adjusting these values can help balance speech length, variation, and synthesis behavior depending on the final use case.

Conclusion

The Voxtral Text to Speech ComfyUI workflow is a compact, practical, and user-friendly solution for generating speech in Videcool. With its combination of a simple node graph, editable speech parameters, and seamless platform integration, it enables beginners and professionals alike to produce narration and spoken audio content quickly and efficiently.